Mercury has been used in gold mining since classical times.  Mercury binds with gold particles giving an amalgam which, due to its density, will lodge in the lining of simple sluice boxes.  This amalgam of gold and mercury, is then burnt and the mercury is released in the form of a volatile gas.  This simple mining process lies behind a complex environmental problem in the Amazon basin, which has important public health implications.

Informal sector mining (garimpagem) has been a feature of the Amazonian economy for at least two centuries.  However, a rise in gold prices in 1979 touched off a major gold rush throughout the Amazon basin, which lasted some fifteen years and particularly affected the Brazilian Amazon.  A combination of falling gold prices and the exhaustion of alluvial deposits had led to a depression in the informal mining sector by the early 1990s, and a significant reduction in the scale of mining activity.  However, mining on some scale continues in most of the traditional mining areas of the Amazon, and the persistence of mercury in Amazonian ecosystems means that the environmental consequences of the mining boom of the 1980s will be felt for many years to come.  Since the mid 1980s, Brazilian researchers and institutions, governmental and non-governmental, have warned of the possible long-term consequences of large-scale mercury use in the Amazon basin.

These can be summarised as follows.  Informal sector gold mining generates two forms of mercury contamination.  The first is caused by inorganic mercury vapours released when mercury/gold amalgam is burnt.  Although the environmental consequences of mercury vapour contamination in the Brazilian Amazon are not yet fully understood, it is certain that it constitutes an occupational health risk for those engaged in mercury burning:  miners themselves, gold traders, and the families of traders, given that burning often takes place within the homes of traders.  The balance of evidence at present is that mercury burning does not constitute a general public health risk for those who live in towns or villages where mercury burning takes place.

The second form of mercury contamination is more complex, and also potentially more worrying, in public health terms.  During the normal course of mining operations mercury is routinely released into the environment.  It may be deliberately thrown into a mining site by miners.  It may be washed out of a sluice box by excessive water flow, or simply spilt while it is being handled.  It invariably ends up in a watercourse.  There it accumulates and is transported down the river system.  After a period it undergoes a chemical transformation, methylation, and becomes organic mercury.  Organic mercury is several orders of magnitude more toxic than inorganic mercury, and has been associated with disastrous mercury contamination incidents, most notably at Minamata in Japan.  Eventually, some of this mercury enters aquatic food chains, contaminating fish.  Since fish is the staple diet of the Amazonian interior, and many Amazonian riverine peasant communities have among the highest per-capita intakes of fish in the world, there is a clear theoretical possibility that mercury contamination in Amazonia could have very serious public health consequences.

Fish contamination with mercury is, however, a complex problem as a result of the biomagnification of this metal in the food chain.  Firstly, this contamination does not affect all species of fish equally:  the research literature clearly shows that carnivorous fish are most likely to be affected, since they absorb the mercury contained in fish they consume , while herbivorous, fruit-eating and bottom-feeding species are less likely to be contaminated.  Secondly, the levels of mercury found even in carnivorous Amazonian fish are not reportedly sufficient to have produced clinical symptoms in adults, although it is possible they may be harmful to children whose mothers were eating contaminated fish after 1979, when large-scale mercury use began.  Thirdly, the extent to which an individual or community might be affected would depend not only on their geographical location in relation to contamination blackspots, but also on the migration patterns of the fish they consume, and the importance of carnivorous species in the local diet.  It should also be remembered that the areas affected are often rural and extremely remote:  no demographic records exist, and no reliable information is available on possible early indicators of a public health problem linked to organic mercury, such as incidence of spontaneous abortion.

Determining the extent to which mercury contamination poses a problem for riverine communities downstream of mining activity is, therefore, a very difficult epidemiological problem.  Even in a severely affected community, not all individuals will be affected:  only those born to mothers living within the community who ate contaminated fish during pregnancy at some time in the 1980s fall within the risk group.  Seasonal variations in fish species consumed might even mean that the field could be narrowed further to those born during the same period to mothers who ate particular fish species at a certain time of year.  Also, there are significant individual clinical variations in relation to mercury exposure, many of which are not fully understood:  in other words, certain individuals with a particular level of mercury contamination might have symptoms, while other individuals with the same level of mercury in their bodies might be completely asymptomatic.

Finally, there is the additional problem that many clinical symptoms associated with mercury contamination are also associated with tropical diseases endemic in Amazonian riverine communities, especially malaria.  Thus even were these symptoms identified within a target group, they might be caused by factors other than mercury  contamination.  These and other complicating factors meant that until this project there was no reliable epidemiological information about the extent of mercury contamination in riverine communities near mining zones, and it was therefore impossible to answer the question of whether mercury contamination was or was not a significant public health problem in Amazonia.  The only certainty was that contamination was occurring.

The origin of most of the mercury contamination is the informal mining sector;  some may be derived from natural sources.  Some pressure has been generated within Brazil to control mercury use by garimpeiros.  Successive federal Mining Codes and state environmental legislation have made the sale and use of mercury in mining illegal.  However, the remoteness of the areas in which informal mining usually takes place, combined with the precarious presence of state institutions in mining zones, means that action against mercury use is almost never taken.  Indeed, the importance of mining to the economy of the interior, even in its current depressed state, often means that environmental legislation is consciously ignored by municipal and even state authorities, anxious to preserve employment and maximise tax revenue.  No practical substitute for mercury in the production process practised by these informal sector miners exists.  The reality of the situation, therefore, is that mercury use in the Brazilian Amazon will continue for as long as informal sector mining remains economic.  Minimising contamination therefore depends on working with miners to modify existing production techniques, rather than on substitution or prohibition.

History of the project

In 1989 the European Commission commissioned a pilot project to assess the extent of mercury contamination, under the direction of Professor Iain Thornton, of IC Consultants Ltd, London and D. David Cleary, then of Cambridge University, to assist with the selection of a field-site and the co-ordination of fieldwork.  After discussions held with members of GEDEBAM, the fieldsite selected was the Tapajós river valley, an area in western Pará state.

The Tapajós valley was selected for a number of reasons.  It is the oldest mining zone in the Brazilian Amazon;  informal sector mining has been continuous in the region since 1956, and on a large scale since 1979.  In terms of both numbers of miners and amount of gold produced the Tapajós has been the most important centre of garimpagem in Brazil since the 1950s, and this is still the case.  The regional centre, the town of Itaituba, is the largest gold trading centre in Brazil, and the whole region underwent an intense mining boom in the 1980s.  At the same time, the region has a significant non-mining population.  A number of small riverine peasant communities are located along the main channel of the Tapajós and its tributaries, totalling around 30,000 people, most of whom rely on fish as the basis of their diet.  Close to the main mining areas is a large Indian reserve belonging to the Mundurucú tribe, some of whom migrate regularly to work in the mines.

All these factors meant that the Tapajós valley was felt to be the logical place to concentrate efforts.  It offered a good range of communities, mining and non-mining, rural and urban, scattered along a river system which extended both north and south of the main mining zones.  It was therefore possible to compare communities close to the sources of contamination with those at a further remove.  In addition, in the Brazilian Amazon the Tapajós has unquestionably been the mining zone where the greatest amount of mercury has been released into the environment over the longest period.  It followed that if mercury contamination really does constitute a significant public health problem in Brazil, it would be in the Tapajós region that the signs would be clearest.

In August and September 1990, a field team assembled by GEDEBAM and working under the direction of ICON, spent six weeks in the Tapajós taking a range of environmental and biological samples from four field locations:  the city of Itaituba, the mines (garimpos) of Cuiú-Cuiú and Creporizão, and the riverine village of Jacareacanga.  126 people donated samples, chosen to represent a number of different risk groups:  fisheating riverine peasants, miners, gold traders, family members of gold traders, and urban residents near trading houses.  Cross-referencing within the sample population also allowed identified control groups for organic and inorganic mercury contamination.

The results showed that some gold traders and their families had levels of mercury high enough to be associated with clinical symptoms of inorganic mercury poisoning, and that some riverine peasants had levels of organic mercury high enough to be associated with clinical symptoms of organic mercury poisoning.  Since no specialised clinical examinations could be carried out during the pilot study, the results were not proof of the existence of clinical symptoms among these risk groups, but rather of their potential exposure.  Two risk groups, such as urban residents near trading posts and miners, were found to have relatively low mercury levels.  The results were set out and analysed in a detailed report submitted to the Secteur Environnement, DG-1 (European Commission) in 1991, “Mercury contamination in the Brazilian Amazon: a pilot project for the Tapajós valley”.  Three recommendations were made:

a environmental and human monitoring should be extended to other communities possibly affected by mercury contamination;
b the link between mercury exposure and health in vulnerable populations should be explored;
c cleaner technologies for gold production and processing should be identified.

After consideration of the report, the European Commission sent a technical mission to Brazil in 1993.  Its brief was to determine whether a broader project was desirable, and, if so, to submit detailed proposals to the Commission.  The mission held a number of meetings with institutions in Brazil to identify potential partners, and identified European sources of expertise in certain highly technical areas where it was felt Brazilian efforts could be strengthened.  The report of the technical mission, submitted in May 1993, proposed a three-year project (eventually extended to four years after evaluation in 1996), which would have five main components:

1 The establishment of field laboratories in the Amazon region for the monitoring and analysis of mercury levels in environmental and human samples;
2 The carrying out of an epidemiological survey to establish the existence and extent of clinical symptoms of mercury poisoning among risk groups in the Tapajós valley;
3 The development and field testing of cleaner mining technologies;
4 A popular education programme to diffuse the results of 2 and 3 above;
5 A regional symposium to disseminate results and lessons to other Amazon basin countries.

These recommendations were accepted by the Commission, and formed the basis for the contract “Mercury contamination from gold mining in the Tapajós and Madeira river basins, Brazilian Amazonia” (ref. no. B7-5041/I/93/15) between DG-1 and Imperial College Consultants (ICON), signed in January 1994. 

Institutional structure of the project

Underlying the specific activities carried out by the project was the long-term goal of strengthening institutional capacity within Brazil, and especially within Amazonia, to deal with mercury contamination.  The institutional structure of the project reflected this aim, with Brazilian institutions being contracted to carry out most of the functions envisaged in the programme of activities, in partnership with European consultants who could bring additional expertise to bear in certain technical areas, where it was felt Brazilian institutions could benefit from external help.  The original proposal was to set up four sub-projects, with the following institutions and external advisors involved:

i  Implantation of dedicated mercury analytical laboratories within the Amazon region, with requisite training of local staff

Objective:  to set up dedicated mercury analysis laboratories working to international standards of quality control within the Amazon region, staffed by local people.

Sub-contractor:  Instituto de Biofísica of the Federal University of Rio de Janeiro, working under the direction of Professor Olaf Malm.  The Biophysics Institute was selected for its distinguished record of research on mercury contamination in Amazonia, and Professor Malm is recognised internationally as a leading authority on the topic.

External consultant and scientific adviser to the programme:  Professor Iain Thornton, Imperial College of Science, Technology & Medicine, London

ii Development of cleaner mining technology appropriate for use in the informal sector

Objective:  to develop mining technologies which will minimise the amount of mercury contamination in the informal mining sector.

Sub-contractor:  SEICOM (Secretaria de Indústria, Comércio e Mineração), an agency of the state government of Pará, under the direction of Sr. Rogério da Silva.  SEICOM was selected for its experience in dealing with the informal mining sector in Pará.

External consultant:  Dr. Hermann Wotruba, Projekt-Consult, Germany

iii Epidemiological survey of groups at risk of mercury contamination in the Tapajós valley

Objective:  to determine the extent of clinical symptoms of mercury contamination among the population of the Tapajós valley.

Sub-contractor:  the Human Ecology Laboratory, Instituto Evandro Chagas, Belém, under the direction of Dr. Elisabeth dos Santos.  The IEC was chosen as being the only regional institution with experience of epidemiological research.

External consultant:  Professor Philippe Grandjean, University of Ødense, Denmark

iv.  Popular health and environmental education campaign in Tapajós valley

Objective:  to raise awareness of the dangers posed by mercury contamination among the mining population, and to publicise the technological modifications developed by the project among miners and mineowners.

Sub-contractor:  GEDEBAM (Grupo para a Defesa dos Ecosistemas do Baixo e Médio Amazonas), under the direction of Sr Pinon Friaes.

External consultant:  none

A local office was to be set up in Santarém at the premises of Fundação Esperança, a Brazilian NGO working on health issues in the Tapajós since the 1970s, which would also be responsible for accounting of local expenditure and disbursement of funds to Brazilian sub-contractors, thus avoiding the bureaucratic complications of channelling the funds through a state organisation. It was envisaged the direction of the project would be shared between a European and a Brazilian co-director.  The European co-director was D. David Cleary.

Laboratories:  Biophysics Institute, Federal University of Rio
Epidemiology:  Evandro Chagas Institute/Grandjean - University of Ødense
Mining technology:  DNPM/Wotruba - Projekt Consult
Education:  AMOT/SEICOM
Project Management:  Paul Docx, ICON
Scientific Adviser:  Iain Thornton, Imperial College

In the absence of a Brazilian co-director, project activities were co-ordinated from the Fundação Esperança office in Santarém by Dr Cleary

Methodology
a  Laboratories

One central problem facing researchers, monitoring bodies and healthcare providers dealing with mercury contamination in the Brazilian Amazon was the absence of local analytical capacity.  Although good quality laboratories for certain kinds of environmental samples existed in Manaus and Belém, these were geographically distant from contamination zones, and could not analyse biological samples, rendering them useless for health investigation.  As a result most mercury samples gathered within Amazonia were analysed outside, in southern Brazil, or as far afield as Europe, North America and Japan.  This posed a series of difficult logistical problems regarding the preservation and transport of samples, which in turn drove up the costs of both sample gathering and analysis, putting it beyond the reach of all but large institutions.  This absence of  reliable sample data, and advice based upon it,  was the main problem facing policymakers and environmental bodies responsible for dealing with mercury contamination in Amazonia.

It was therefore felt that by setting up dedicated mercury analysis laboratories within the Amazon region the project could make an important contribution to strengthening local technical and institutional capacity to deal with the mercury problem.  But the problems facing an attempt of this kind were considerable.  Firstly, buying and installing sophisticated analytical equipment on its own would not solve anything, if local people could not operate them.  A long-term training programme of locally recruited laboratory staff would clearly be as important as the physical establishment of a laboratory.  Secondly, the equipment would need to be sufficiently robust to withstand the difficult operating conditions it would encounter in the Amazon, where power cuts are common.  Thirdly, the equipment would need to be repairable locally, as far as possible;  while in-country experts could give advice down phone lines from southern Brazil, local laboratory technicians needed to be able to diagnose problems and improvise solutions themselves.  Finally, in the long-term the laboratories needed to generate some income of their own.  Besides reducing their dependence on external funding, it would also increase the value of the laboratories to the local universities which would receive them as a donation at the end of the project.

It was decided to locate two laboratories within the Amazon, at Santarém (Pará state) and Porto Velho (Rondônia state). 

Both sites were visited by Professors Malm and Thornton during the technical mission, and detailed plans were drawn up in 1993 for the conversion of the laboratory areas to receive equipment.  It was decided that the laboratories should divide the analytical requirement of the project between them, with Porto Velho specialising in environmental samples, and Santarém in biological samples.  Both laboratories, should be capable of carrying out analysis of fish samples.

The analytical equipment selected for both laboratories was a Perkin-Elmer FIMS atomic absorption spectrophotometer for mercury analysis, with Windows-based software, and a microwave unit for sample digestion.  It was not felt that speciation between organic and inorganic mercury was necessary, given that mercury from urine samples could be safely assumed to be mainly inorganic, and mercury from fish and hair samples would be mainly organic.  The advantage of the microwave unit and the computer control of the spectrophotometer was felt to be that it automated many aspects of sample preparation and analysis which would otherwise have to be performed by laboratory staff.  In a situation such as that encountered in Santarém and Porto Velho, where no laboratory staff could be expected to have had experience in heavy metal analysis, this was felt to simplify training requirements, and thus increase the number of people who could be trained in laboratory techniques during the life of the project.

b.  Epidemiology

The project concentrated on two risk groups:  gold traders at risk of inorganic mercury contamination through amalgam burning, and riverine peasants at risk of organic mercury contamination through fish consumption.  The first group was readily identifiable:  gold trading posts are easily mapped walking through an urban area, and a list can be made of those working in them.  The symptomology of vapour contamination is relatively straightforward, and has been thoroughly documented in the international scientific literature as a result of contamination incidents in the industrialised world.  It was therefore not difficult to draw up a number of questionnaires and examination protocols looking at the medical history, current clinical condition and occupational record of those working in gold trading posts.  Carrying out an epidemiological investigation of mercury contamination in gold trading posts is therefore not a difficult methodological exercise, provided that the preliminary field survey was properly done, and the logistical problem of transporting the examinees to the field station could be dealt with.

The following methods were used in the epidemiological investigation of employees in gold trading posts.  First, a survey was carried out to identify the sample population.  In accordance with Brazilian legislation, the objectives of the research were explained and signed consent forms were obtained from those who agreed to participate.  On the designated day and time, examinees were taken to a field station, where the following examinations and questionnaires were applied:

a A general clinical examination
b A medical history questionnaire
c An occupational history questionnaire
d A specialised clinical examination and questionnaire for organic mercury contamination
e An examination of early indicators of contamination (grip strength, reaction time and co-ordination) to assess dose/response relationships.

The results were entered in a databank for statistical interrogation, using standard programmes (Epi-Info and Dbase).  A control group selected to mirror the age and sex profile of the employees was also examined, for comparison.

In methodological terms, the investigation of organic mercury contamination among fish-eating riverine communities is much more difficult.  The following problems were encountered and dealt with:

a.  Lack of trust within communities:  It was a natural assumption on the part of the communities involved that the arrival of a research team either indicated the existence of a severe problem which was being shielded from them, or had an ulterior political motive.  Weeks before the arrival of the research team, a delegation from the project met community leaders to explain the rationale behind the work, and visited every household.  During the fieldwork a number of health professionals in the investigation team were assigned to providing general health care to members of the community free of charge, whether or not they were part of the investigation.  Afterwards, every individual examined (several hundred in total) was sent a personal letter by the project director and the Evandro Chagas Institute giving the results of all samples and making recommendations as appropriate.

b.  Lack of baseline demographic information:  A detailed census and household survey was carried out in the preliminary visit.

c.  Illiteracy:  Care was taken to select diagnostic tests which relied upon visual and physical skills, rather than literacy.  All tests were vetted by an anthropologist, to ensure that they were culturally appropriate.

d.  Language:  Inhabitants of riverine villages in the Tapajós speak a rural dialect Portuguese.  For an accurate response to questionnaires it is important that it should be framed using language that is comprehensible to the examinee rather than the examiner.  Questionnaires were therefore vetted by a local doctor in Santarém thoroughly familiar with the dialect.  Particular challenges were posed by Indian villages, where many inhabitants are still monolingual in Mundurucú.  In the one Indian village investigated interpreters were used, and certain diagnostic tests felt to be culturally inappropriate were dropped.

e.  Geographical location:  The remoteness of the villages posed a particular logistical challenge.  There was often no accommodation available for the research team, no power supply, and no water supply.  In most cases, the research team stayed on a boat, hired for the duration of the stay.  A portable generator was used, together with a motor pump.

The selection of communities to be investigated was determined by a number of variables.  The most important was size:  the research team could only stay in the field for a limited amount of time, and process a limited number of people.  This number needed to be a statistically significant percentage of the defined risk group within the population.  In effect, this meant restricting investigations to communities of between 500 and 1,000 inhabitants (in terms of the Tapajós valley, a medium-sized village).  Geographical location was also important.  It was necessary to find a control community where mercury contamination was known not to be a problem, and it was also necessary to gather information from communities which would be representative of the situation along different stretches of the river valley, relatively close to the contamination zone and further away from it.  The communities selected for investigation were the villages of  Sai-Cinzas (upper Tapajós valley close to mining area:  Mundurucú Indian community within the Mundurucú reserve);  São Luis do Tapajós (Brazilian peasant village with some Mundurucú inhabitants, upper Tapajós valley but further away from contamination zone);  Brasília Legal (exclusively Brazilian peasant village, middle Tapajós valley);  Santana do Ituquí (exclusively Brazilian peasant village on the Amazon east of Santarém, control group).  All these villages are identified on Map 2.

Not all inhabitants of riverine villages were regarded as at-risk from organic mercury contamination.  Within communities investigation centred on two groups: women aged 15-49 and children aged 7-12.  Children born after 1979 and living continuously in the community were seen as the most important risk group; the 7-12 age range was selected as a consequence of the particular diagnostic tests chosen, which required a child to understand and carry out instructions.  Women in the fertile age-range were of interest as potential mothers: given the possibility that mercury might cause problems in utero, the health history and mercury levels of potential mothers is important contextual information for assessing contamination in children.

The selection of these risk groups was central to the project’s work in this area.  It was felt important to narrow the focus of attention to those with high potential exposure, and those with a possible hypersusceptibility to toxic effects.  Given the special sensitivity of the foetus, the exposure levels of women in childbearing age are of particular interest, and adverse health effects occurring in a population with chronic methylmercury exposure would be seen first in those exposed in utero.  Since the neurobehavioural tests that can be administered to infants are relatively crude, studies of this age group probably would not reveal slight toxicity.  More sensitive methods can be applied from ages about seven years and up.

The following information was gathered:

Women aged 15-49

a General clinical examination;
b Medical history questionnaire, concentrating on pregnancies and child development;
c Hair sample for mercury analysis;
d Blood, stool and urine samples for general analysis.

Children aged 7-12, workers in gold trading posts

a General clinical examination;
b Medical history and developmental questionnaire (from mothers or other close relative);
c Specialised examination for mercury contamination;
d Hair sample for mercury analysis;
e Blood, stool and urine samples for general analysis;
f Series of diagnostic tests for mercury contamination.

The diagnostic tests chosen were simple to apply and, as far as possible, culturally neutral:  they examined attention span, short-term memory, mood, and visuospatial/visuomotor performance.  The tests chosen were Stanford-Binet Bead Memory Test, Copying Test, Copying Blocks Test, WISC Span Digits Test, and Boston Cartoon Mood Test.  Since datasets exist for these tests from a number of populations around the world, they do allow international comparisons to be made.

c.  Mining technology

Three basic assumptions were made by the project.  The first was that it was not possible to treat mercury recovery in isolation from prospecting and gold production processes.  The second was that the only way to ensure take-up of cleaner technologies would be to associate changes in technique with increased gold production, creating a direct economic incentive for the miner.  The third was that it was essential to modify existing techniques, rather than import new technologies from outside the garimpo.

The methodological implication was that the project would need to modify prospecting and mining techniques in order to attack the mercury problem.  It was not felt that concentrating on mercury use to the exclusion of other mining habits would offer sufficient incentive for take-up.  The aim was rather to offer an integrated set of techniques to the miner which would minimise a range of environmental impacts of informal sector mining, from deforestation caused by traditional prospecting through sedimentation of watercourses to degradation caused by inefficient excavation and disposal of tailings.  Although not always logically dependent on one another, they would be associated in the minds of miners and mineowners with increased gold production and minimal use of mercury.

The method developed for rationalising prospection techniques was to teach miners to take samples at regular intervals from transects across locales selected by the miners themselves on the basis of local geological knowledge.  A simple bench core-sampler (sonda de banca), already widely used by garimpeiros in the Tapajós, was used to take the samples.  These were panned, dried, and examined under a simple binocular microscope by a trained garimpeiro.  Gold grains were counted and graded, and a miner was trained to calculate a production estimate on the basis of factors of production, using standard statistical techniques.  In theory, this would allow mining machinery to be deployed only in areas where sufficient gold existed to make its use economic, avoiding the unnecessary environmental destruction that often resulted from traditional prospecting techniques, which often led to excavations taking place in areas where little or no gold was found.  However, mining populations are conservative and to convince them it was necessary to compare production estimates with real production figures from mining operations carried out by the project.

Much mercury contamination in garimpos can be attributed to spillages during the production process.  Most of this spillage occurs from sluice boxes, where mercury is introduced at the end of the mining process to amalgamate with gold particles.